Artificial Intelligence Design of Antimicrobial Nanomaterials: Prediction, Optimization, And Discovery

Abstract

Antimicrobial activity of functionalized nanomaterials is difficult to predict due to performance that varies according to composition, surface chemistry, biological media and microbial context. Progress in artificial intelligence and data-driven modeling is making screening and optimization of nanomaterial designs possible, through learning the relationship between nanomaterial descriptors and antimicrobial responses. The present paper provides an overview of the evidence on the design of artificial intelligence-enabled nanomaterials, nanotoxicology modeling, and translational antimicrobial approaches to suggest a reproducible workflow to predict antimicrobial activity of functionalized nanomaterials. Evidence was visualized over a lifecycle spanning from the definition and curation of the descriptors, training between validation, and deployment that include translation-awareness. In the literature, there are strengths such as rapid iteration and design-space exploration, and repeated issues of limited standardized datasets, heterogeneous protocols of measurement, bias and lack of external validation. Through matching the predictions of artificial intelligence and the transparent data practice and stage-resolved validation, this framework can enhance the more believable, comparative, and secure development of antimicrobial functionalized nanomaterials.